ENVS 4432 Assignment 6

Github user name: SophieKent-Purcell

A Summary of: Landscape connectivity for bobcat (Lynx rufus) and lynx (Lynx canadensis) in the Northeastern United States

Farrell, L. E., Levy, D. M., Donovan, T., Mickey, R., Howard, A., Vashon, J., Freeman, M., Royar, K., & Kilpatrick, C. W. (2018). Landscape connectivity for bobcat (Lynx rufus) and lynx (Lynx canadensis) in the Northeastern United States. PLoS ONE, 13(3), e0194243. https://doi.org/10.1371/journal.pone.0194243

Introduction

Habitat fragmentation has been increasing across the Northern United States due to human expansion. This is problematic for wildlife species such as bobcat (Lynx rufus) and lynx (Lynx canadensis), which require large amount of connected habitat for their seasonal shifts in habitat and for long term gene flow within metapopulations. Creating conservation strategies to ensure species have adequate connected habitat is challenging because different species like bobcat and lynx have largely different habitat requirements and different spatial scale for these requirements. An additional challenge is presented by the fact that in many areas, the suitability and connectivity of habitat for many species is unknown. GPS tracking is one way to identify what conditions are required for suitable habitat. This information can then be used to model how core habitat areas are connected. Such models are lacking for both lynx and bobcat in the Northeast United States. This study aims create to create a habitat suitability model for lynx and bobcat across Vermont, New Hampshire, and Maine based on habitat suitability requirements determined from a Partitioned Mahalanobis Distance Squared analysis (D²) using GPS collar data. Additionally, it aims to model current connectivity across subportions of these states and estimate how this connectivity will change with future increases in anthropogenic pressure.

Methods

The area being analysed in this study covers the states of Vermont, New Hampshire, and Maine (Fig 1.). GPS radio-collar location data for lynx were provided by the Maine Department of Inland Fisheries and Wildlife and consisted of data from 2005 to 2010. Similarly, bobcat GPS radio-collar locations from 2005 to 2008 were provided by the Vermont cooperative Fish and Wildlife Research Unit. A maximum of 200 locations was chosen from each collar to minimize bias from collars used for longer. To assess habitat suitability, lynx and bobcat occurance data were compared to 39 environmental variables at three spatial scales; local distance, daily distance, and female home range size. A neighborhood analysis was preformed on binary layers representing each variable to determine which variables were consistently present across species occurrence locations, with a consistent variables being one with a coefficient of variation <= 1. This and all other mapping operations were done in ArcMap 9.3 or ArcMap 10. After this, a principal components analysis (PCA) was completed using all the consistent variables in order to identify variables to use in the final model. The results from this were used in a D² analysis for each species to identify the suitability of each 30m pixel of land area. The resulting D² values were then transformed into P-values, which in this case represent how close an area is to ideal habitat. Five subportions of the regional analysis area, 3 for bobcat and 2 for lynx, were selected for connectivity analysis based on their similarity to the areas where radio-collar data was collected. This connectivity analysis was done based on the values from the D² analysis using Linkage Mapper v6 by identifying least cost corridors between core areas, represented in this study by land conserved either privately or by the federal or state government. Finally, the persistence of connectivity was estimated by using the projected change in human footprint to identify how the cost of each connectivity corridor may change in the next 30 years.

Figure 1: Regional area of analyses covering the states of Vermont, New Hampshire, and Maine. Included are the study areas where GPS radio-collar locations are from. Basemap shade indicates habitat suitability as found in this study with lighter areas being more suitable.

Results and Conclusion

Originally, 39 habitat variables were identified for the use in the study. Of these, 31 and 35 were deemed to be consistent for bobcat and lynx respectively and were carried forward. Using combinations of these variables, 68 and 84 bobcat and lynx PCA models were created. From these models, the 5 variables that accounted for the greatest variation were chosen to be used for connectivity analysis. For bobcat these were ecotone/edge density, undeveloped habitat, water within 300m, cover within 300m, and cover edge 300 m margin. For lynx the variables were shrub scrub, natural habitat, cover within 300 m, elevation, and slope. The PCA also showed that lynx have more specialization of habitat than bobcat. Connective habitat for bobcat accounted for 96.5%, 96.6% and 69.3% of the area within the 3 subportions analysed. For lynx, these values were higher with connective area covering 98.3% and 98.2% of the 2 subportions. By the year 2040, connective bobcat was estimated to decrease by 16.9%, 11.9%, and 6.7% in each subpoption due to increased human footprint. The impact on lynx connective habitat was estimated to be smaller, with decreases of 3.4% and 0.3% in the subportions. The authors were able to achieve their aims of modelling habitat suitability, current connectivity, and future changes to connectivity based on estimated increases in anthropogenic pressure. Connectivity work, such as that done in this study is important for understanding how land management decisions will impact the ability of wide ranging species to travel through the landscape.